AlphaZero sheds new light on the grand game of chess

Photos by: Luiz Hanfilaque on Unsplash

Google’s DeepMind (behind AlphaGo) just generalized a single AlphaZero model to master: Go, chess, and shogi. AlphaZero is self-taught and is becoming the best player in history for each of these, despite having n0 domain knowledge. It starts its training from random play with an untrained neural network.